The Rangers `Sophisticated` Stats Package

In a recent interview with Newsday, Rangers coach Alain Vigneault dropped this quote:

 Newsday: Ryan McDonagh returning to form for Rangers | 12/20/2017

Newsday: Ryan McDonagh returning to form for Rangers | 12/20/2017

Now, in AV's defense here, when the team you coach is 30th in the league in shot attempts against per 60 minutes of 5v5 play, well, you can't really go on record saying that you care about it or coach it because, well, you're terrible, and that wouldn't be a good look for you. 

But this isn't really a good look for the Rangers, either. The studies are essentially conclusive at this point. Shot attempts, Corsi, is the foundation of the analytics movement. CF% is a better predictor for future team success than scoring chances, some xG models, and actual goals. These are all things we know, and the debate on these should be behind us at this point. Then, you see a quote like this from a respected coach in the NHL, and we get set back.

Let's see if we can figure out what the Rangers do value in their `sophisticated` stats package, and let's see if we can do it using only the data available on Corsica.Hockey, publicly available data. Obviously we won't know for sure, but let's see what we can infer.

The picture we'll be looking at is 5v5 data, unadjusted, team ranks for this season. We already know that the team is 31st in CA/60, but in some other telling metrics, they rank:

26th in CF/60
30th in CF%
7th in GF/60
21st in GA/60
15th in GF%
2nd in xGF/60
31st in xGA/60
20th in xGF%
1st in xFSh%
30th in xFSv%

Above, I've bolded the highlights. 

What can we infer? The Rangers don't care about quantity, they care about quality. But, the damage they're doing to themselves on defense with the quality against they allow is partly cancelling out the good things they're doing on offense to generate the quality shots. Can one exist without the other?

I think this is AV hockey in a nutshell. Sacrificing defense for offense. The man-to-man scheme that the Rangers employ in the defensive zone allows for quick breakouts, rush opportunities, odd-man rushes. But, it leaves Lundqvist and Pavelec out to dry, and if they aren't making the saves, and the Rangers aren't finishing their chances, the score gets ugly.

Luckily for the Rangers, their goaltending duo has been carrying the load and masking a lot of the defensive deficincies we see with our eyes and on the stat-sheet.

Sean Tierney, ChartingHockey, has a very handy-viz based off of Manny's K metric that displays the Rangers thought process quite well:

 The visualization displays what is discussed above. The Rangers strength lies in the quality they can generate for themselves. The Rangers weakness lies in the quality against they allow. The system?

The visualization displays what is discussed above. The Rangers strength lies in the quality they can generate for themselves. The Rangers weakness lies in the quality against they allow. The system?

 

Further, we need to debate sustainability. What's more repeatable, quantity or quality? If you're coaching to quality, as the Rangers clearly are, are you playing with fire?

To isolate what teams are generating in terms of quantity versus quality, we'll be exploring split CF/60 (a measure of quantity generated) and split xFSh% (an isolated quality metric based off of Manny Elk's expected goals model). I'll also be using a split-half of last season's games in order to get more data.

First half: through Jan 10th

split_cf60.png
split_xfsh.png

On the visualizations above, the blue line represents the line of best fit via a linear model, the gray line is a simple x=y line.

Basically, the visualizations show us what we know, what we've known since the beginning of the movement. Shot attempts have a very strong repeatability. It's a skill. We don't see the same strength in shot quality. 

Flaws in the above: The sample size is way, way too small. To make a stronger point, we should analyze and chart these splits for each season going back to 0708. I didn't do that. I should've done that.

Thus, if you're coaching to quality rather than quantity (or a strong mix of both, which is pretty clearly the best option), then you're playing with fire. This `sophisticated` stat package that the Rangers are boasting essentially seems to boil down to what the Rangers thought they were getting when they originally hired John Tortorella, safe is death.

Now, 35 year-old Lundqvist is still producing as one of the best goalies in the league (8th in the league in goals saved above expected per 60 among starters), and the system and stats-package that the Rangers and AV are utilizing is exploiting that fact. And maybe they should be, but there is a serious lack of quantity, and we see that relying on quality isn't as sustainable.

Further, noting again how reproducible quantity is, we see the Rangers being nearly dead last in the league in terms of allowing attempts against. Unless the system changes, well, there's no end in sight for this.

We can debate endlessly the types of players that we think the Rangers are missing, specifically in my opinion, a RHD to pair with McDonagh that is an undervalued shot suppressor (think Connor Carrick in Toronto, Josh Manson in Anaheim before people woke up, etc...), and perhaps maybe a winger in the mold of Benoit Pouliot circa 1314 to complement our centers and make sure the game is going in the right direction. But if the system isn't going to change, or if we're going to bring these types of players into the fold just to find Carrick in the pressbox instead of Steve Kampfer, then what's the point?

The `sophisticated` package and system are played out. We see it every single year. The Rangers launch themselves into the playoffs via a ridiculous PDO, and then the game tightens up, the refs swallow their whistles, there's less space on the ice, there is less opportunity for rush attempts for. The game clogs. And what do we see in the playoffs? Lundqvist left out to dry, because that's the system, and the offense disappears.

Rinse and repeat, but `sophisticated`.